Traditional Observability Wasn’t Built for AI Failures | Shahar Azulay, groundcover | TFiR
AI systems fail differently than traditional applications—no error codes, just subtle drift and hallucinations that sampling will ...
Your AI Storage Bill Is Lying to You — Stefaan Vervaet, Akave, Is Fixing That | TFiR
Akave Founder and CEO Stefaan Vervaet joins TFiR to explain how their decentralized, S3-compatible storage platform eliminates ...
AI Infrastructure Lock-In: Why PyTorch is the Only Abstraction Layer That Matters | Mark Collier | TFiR
Enterprise AI infrastructure faces a fragmentation crisis as hardware and software evolve simultaneously at unprecedented speed. Mark ...
AWS Testing Costs Kill Velocity: LocalStack Hits 400M Docker Pulls | TFiR
Cloud development is broken. LocalStack crossed 400M Docker pulls by running AWS entirely on laptops—eliminating deployment friction ...
Real-Time Streaming Unlocks Agentic AI at the Edge | Prenil Kottayankandy, Akamai | TFiR
Prenil Kottayankandy of Akamai explains how the Akamai x Redpanda partnership unlocks the real-time streaming data ...
Three AI Bottlenecks That Will Break Enterprise Architectures in 2026 | Danielle Cook, Akamai | TFiR
Danielle Cook, Senior Manager at Akamai, breaks down the three enterprise AI bottlenecks for 2026 — centralized ...
Hybrid Cloud Is Breaking Streaming Architecture—Here’s How to Unify It | Prenil Kottayankandy, Akamai & Zeke Dean, Redpanda | TFiR
Hybrid deployments create streaming silos and data consistency problems. Portable streaming infrastructure with agentic data plane controls ...
Autonomous Workers Replace IT Agents for L3/L4 Tickets | Nenshad Bardoliwalla, ServiceNow | TFiR
ServiceNow's autonomous workers operate with a user ID, pull from the same queue as human agents, and ...
AI Pilots Are Failing Everywhere. Scott Morgan of Marlabs Says Governance Is the Fix | TFiR
Marlabs EVP Scott Morgan breaks down why AI pilots stall before reaching production and how the ...
GPU Partitioning and Scheduling: Complementary Approaches to Efficient GPU Utilization
Modern GPUs deliver enormous compute power, but many workloads don't require an entire device. GPU partitioning technologies ...